# Sail Research vs Luminal

> Sail sells slow inference at 30–80% off for patient workloads. Luminal sells faster engines that cut cost per token in real time.

Canonical: https://www.subconscious.dev/compare/sail-research-vs-luminal · By The Subconscious Team · Updated September 30, 2026

## How they compare

Sail Research prices by completion window: wait minutes per turn and save 30% to 80% on open models like Kimi K2.6 and GLM-5. It is built for jobs that can wait. Luminal cuts cost the other way, by compiling a model into fused native kernels ahead of time so each GPU serves more tokens, reporting 36K per second on GPT-OSS 120B over 8 H100s.

Sail suits background agents and batch jobs where latency does not matter. Luminal suits teams serving a model in real time who want a cheaper engine, or who need it on-prem. Luminal's prices are not public, so compare quotes against Sail's discounts directly.

## What each one does

### Sail Research

Sail Research sells throughput over latency. Founders Neil Movva and Samir Menon built a serving stack that packs as much work as possible into every GPU, and customers state how long they can wait through completion windows. The priority window targets about a one-minute turn for roughly 30 to 50% off the immediate asap price. The default standard window targets about five minutes for 45 to 65% off. The flex window runs off-peak for 60 to 80% off.

### Luminal

Luminal builds an inference compiler. Where vLLM and SGLang interpret a model at runtime, Luminal compiles it ahead of time into native kernels for GPUs and ASICs. Models get lowered to a small graph of 15 primitive ops, and the compiler searches over fusion, tiling, memory and scheduling choices instead of relying on hand-written rules, which it says can find optimizations like Flash Attention on its own. The compiler is open source in Rust under Apache 2.0 or MIT, runs on CUDA and Metal with ROCm on the roadmap, and works as a torch.compile backend.

## Which is best, and when

### Choose Sail Research for

- Background agents that can wait minutes
- Deep discounts on open models
- Customer LoRA fine-tunes

### Choose Luminal for

- Lower cost without giving up real-time latency
- On-prem deployments with custom kernel work and SLAs
- Serving custom or fine-tuned architectures off any catalog

## At a glance

| Attribute | Sail Research | Luminal |
|---|---|---|
| Model access | Open weights | Bring your own weights |
| Flagship models | Kimi K2.6, GLM-5, GPT-OSS 120B | No public catalog |
| Speed | Minutes per turn by design | 36K tok/s on GPT-OSS 120B, 8xH100 (vendor) |
| Price | 30–80% off by completion window | Pay per use; rates not published |
| Customization | Customer LoRA fine-tunes | Compiles any PyTorch or HF model |
| Deployment | API plus Sailboxes | Serverless (early access), on-prem license |
| Long context | Varies by model | - |

## FAQ

### What is the difference between Sail Research and Luminal?

Sail sells slow inference at 30–80% off for patient workloads. Luminal sells faster engines that cut cost per token in real time.

### When should I choose Sail Research over Luminal?

Background agents that can wait minutes; Deep discounts on open models; Customer LoRA fine-tunes.

### When should I choose Luminal over Sail Research?

Lower cost without giving up real-time latency; On-prem deployments with custom kernel work and SLAs; Serving custom or fine-tuned architectures off any catalog.

### Is Sail Research or Luminal cheaper?

Sail Research: 30–80% off by completion window. Luminal: Pay per use; rates not published. The cheaper choice depends on the model and workload.

## Running long-horizon agents?

If your agents run past 200K tokens, compare both against Subconscious: [Subconscious vs Sail Research](https://www.subconscious.dev/compare/subconscious-vs-sail-research.md), [Subconscious vs Luminal](https://www.subconscious.dev/compare/subconscious-vs-luminal.md).

Full profiles: [Sail Research](https://www.subconscious.dev/providers/sail-research.md), [Luminal](https://www.subconscious.dev/providers/luminal.md).